Consistency: Why Must Multiple Copies Follow Shared Rules?

Consistency does not mean that every copy must be identical at every instant. It specifies the relationships that their reads and updates must obey. A shared calendar keeps data on both a phone and a computer. Just after you change a meeting time, briefly seeing the old time is not necessarily an error, but the system must state when the update will propagate and whether later reads will reflect a confirmed change.

Consistency is related to synchronisation but is not the same thing. Synchronisation concerns how separate processes align in time; consistency concerns whether different observations still satisfy shared rules. Nor is it accuracy: two copies may agree with each other while both contain the wrong venue.

Stronger consistency is therefore not automatically better. Making every location show the same result immediately can cost time, communication and availability. Allowing controlled temporary differences may make a system faster and more tolerant of disconnection. The real task is to specify which relationships must never be broken and how long any permitted difference may last.